ops-deploy

name	ops-deploy
description	Deployment, CI/CD, monitoring, and production operations for RunSmart PWA. Covers Vercel deployment, environment management, performance monitoring, error tracking, and production incident response. Use when deploying, configuring infrastructure, or troubleshooting production issues.
metadata	{"short-description":"Deploy, monitor, and operate RunSmart in production on Vercel."}

When Claude should use this skill

• Deploying RunSmart to production or preview environments
• Setting up or debugging CI/CD pipelines
• Configuring environment variables across environments
• Monitoring performance, errors, or uptime
• Responding to production incidents
• Optimizing build times or bundle size

Deployment Stack

Layer	Service	Purpose
Hosting	Vercel	Next.js deployment, edge functions
Database	IndexedDB (client)	Local-first data persistence
AI API	OpenAI via Vercel AI SDK	Chat, plan generation
Analytics	Vercel Analytics / Web Vitals	Performance monitoring
Error Tracking	Console + Vercel Logs	Error reporting
PWA	next-pwa / Service Worker	Offline support, install prompt

Deployment Checklist

Pre-Deploy

1. npm run lint — zero warnings
2. npx tsc --noEmit — zero type errors
3. npm run test -- --run — all tests pass
4. npm run build — clean production build
5. Verify environment variables are set in Vercel dashboard
6. Check bundle size: npx next build --analyze (if configured)

Deploy

# Preview deployment (PR-based)
git push origin feature/branch-name
# Vercel auto-deploys preview

# Production deployment
git push origin main
# Vercel auto-deploys to production

Post-Deploy

1. Verify production URL loads correctly
2. Test PWA install flow on mobile
3. Check API routes respond (health check)
4. Verify AI features work (chat, plan generation)
5. Monitor error logs for 30 minutes

Environment Variables

# Required for production
OPENAI_API_KEY=sk-...          # AI features
NEXT_PUBLIC_APP_URL=https://runsmart.app

# Optional
VERCEL_ANALYTICS_ID=...        # Vercel Analytics
SENTRY_DSN=...                 # Error tracking (if added)

PWA Configuration

• Service Worker: caches app shell, static assets, API responses
• Manifest: public/manifest.json with app name, icons, theme
• Offline: Core features work offline via IndexedDB
• Update: Background SW update with user prompt to refresh

Performance Targets

Metric	Target	Tool
LCP	< 2.5s	Lighthouse
FID	< 100ms	Web Vitals
CLS	< 0.1	Web Vitals
TTI	< 3.5s	Lighthouse
Bundle (JS)	< 200KB gzip	next build
PWA Score	100	Lighthouse

Incident Response

1. Detect: Vercel logs, user reports, monitoring alerts
2. Triage: Classify severity (P0-critical to P3-cosmetic)
3. Mitigate: Rollback via Vercel instant rollback if needed
4. Fix: Hot-fix branch → PR → merge → auto-deploy
5. Review: Post-incident review, update monitoring

Agent Team Pattern: Release Swarm

Create an agent team for production release:
- Build agent: full production build, fix any errors
- Test agent: run all test suites, report failures
- Deploy agent: verify Vercel config, environment vars, deploy
- Monitor agent: check post-deploy health, logs, performance

Integration Points

• Build: V0/next.config.ts, V0/package.json
• PWA: V0/public/manifest.json, service worker config
• Env: .env.local, Vercel dashboard
• CI: GitHub Actions (if configured), Vercel build hooks